DC-DC boost converter problem

[adeel abid]

Dear All,

i have designed a boost converter. It is based on IRF740. Driving gate signal is generated by microcontroller which is then pulled up with !k resistor to 12V. So Vgs is either 12V or 0V with ton=190us and min. toff=10us giving about 93.3% duty cycle.The gate driving signal is not continuous but actually based upon monostable multivibrator that is only trigerred if load voltage is less than desired value & simultanoeusly indcutor current is reduced to zero (this is sensed by sensing voltage across 1ohm resistor).

Problem i am facing is that when MOSFET is on for 190us as per ideal calculation id should be ramp with ID(max)=8A. During this phase only the battery, inductor and IRF740 is in the circuit. But practically after implementation i am getting ID(max)=3.5 to 4A (measured by sensing peak voltage acorss 1 ohm resistor).

There is not so much heating of IRF740 to increase RDs(on) to limit the drain current to this value because i have tested it with no load i.e. without 500 ohm resistance (and hence multivibrator pulse repetition period is quite large to allow cooling between generation of pulses). This is limiting the input power of boost converter and hence the output power delivered to the load.

please provide reason for this behaviour.

Thanks in advance.

 

[BradtheRad]

The switching device must have very low resistance, or it drastically limits power output.

In my comparisons using simulations, I find when I install a transistor or mosfet, they have significant On-resistance. Less current can build through the coil. It reaches a limit even if I leave it conducting.

However when I install an artificial switch, I can obtain more current, because I can reduce On-resistance to a tiny amount. Flow builds quickly and keeps building.

Here is a screenshot comparing two boost converters. The left has 1/10 ohm switch resistance.

The right has enough ohms added in series, so as to match the coil current you stated. The amount needed to match is 2.5 ohms.



Since half the current can flow in the right-hand layout, only half the output voltage is generated on the load.

The 5k resistors are included so the simulator won't generate false oscillations at idle.

At first I thought your 1 ohm resistor was to blame, however a simulation showed it only causes a slight reduction.

What will help? Anything to drive the mosfet to have less On-resistance. You may even have to put two or more in parallel.

 

[FvM]

Problem i am facing is that when MOSFET is on for 190us as per ideal calculation id should be ramp with ID(max)=8A.

We usually find ideal calculations in the land of unrealistic design assumptions where real world component parameters don't exist.

A realistic calculation should a least consider Rdson and inductor series resistance. So you clearly won't achieve 8 A. As a trivial point, inductance may be higher than nominal value.

I also don't understand how the gate drive circuit looks like. Depending on the actual gate waveform, there may be switching losses that make IL drop before the current commutates to the secondary. An accurate measurement should be performed in the on-phase. I see that it needs a more sophisticated setup, but it's worth the effort.

 

[adeel abid]

Depending on the actual gate waveform, there may be switching losses that make IL drop before the current commutates to the secondary. An accurate measurement should be performed in the on-phase. I see that it needs a more sophisticated setup, but it's worth the effort.

please find the attached figure
upper trace is of gate waveform and lower trace is of voltage across 1ohm resistance.

may be this can help u to find out real problem

 

[FvM]

Gate drive looks good, I guess you have 12V Vgs? Inductor resistance should be known, also are you sure there's no input voltage when IL rises?

 

[BradtheRad]

Your scope image looks normal.

However...

You are getting only half the picture.

Since the coil is the center of action, you'll get more information by taking scope readings of current through the coil. Your low resistance should be put in line next to the coil. That way you can examine the entire power cycle on the scope.

A 1/10 ohm resistor should be low enough. You don't want to reduce current flow unnecessarily.

Notice my right-hand simulation (above) shows a much different curve (as compared to the left-hand). It is obvious that it will reach a plateau level of approximately 4 A. This portrays a faster inductive time constant, calculated as L/R.

 

[adeel abid]

please find the picture for inductor current (1A per div ,easurd with 0.1ohm in series with inductor) and inductor voltage waveform.

please advise

thanks in advance

 

[BradtheRad]

1.

Yes, that tells a lot. The trace does not look faulty.

Current builds steadily. It has not started leveling off at the peak. It could continue to build.

However it gets switched off.

Looking at your initial schematic, it appears the frequency is 5 kHz. If you could slow this down, there would be more time for current to build through the coil.

My simulation shows you would get 5A peak through the coil if the frequency were 3.4 kHz. You stated the frequency comes from a regulating method, so you may need to override this in order to achieve a lower frequency.

2.

Another thing that could help is to add one or more mosfets in parallel. It should reduce overall resistance during On-time. This may require a lot of work to get them adjusted so they match in performance.

3.

Here is another idea. Twin interleaved boost converters. It works in theory.



The clocks must alternate. Their duty cycle each can be 93%. This will draw more current from the battery, but the peak (6A) is still lower than the 8 amps you thought you would get.

I can't say how easy it will be to make this arrangement work. The simulation doesn't show harmful effects.

 

[FvM]

Your current waveform shows, that the inductance is apparently larger than 300µH, as already assumed in post #3. To be sure, you'll want to check that the 12 V supply doesn't drop during on phase.

 

[adeel abid]

voltage is controlled by changing turn off time of mosfet. turn on time is fixed and is 190us. considering Rds(on) to be 0.5ohm, simulation gives peak of 6.5A but i am not getting it. any possible reason for it?
also note that current rises with higher rate initially but rise rate slowed down and continued for rest of time.
please tell what is stopping it to get peak current of 6.5A?
Thanks in advance.

- - - Updated - - -

Your current waveform shows, that the inductance is apparently larger than 300µH, as already assumed in post #3. To be sure, you'll want to check that the 12 V supply doesn't drop during on phase.

there is a voltage drop of not more than 0.5V during turn on period. there are ripples but i think they are not significant.
any suggession to reduce it further?
thanks in advance

 

[FvM]

also note that current rises with higher rate initially but rise rate slowed down and continued for rest of time.
please tell what is stopping it to get peak current of 6.5A?

If you verified that the effect is not caused by a supply voltage drop, I would expect that the final dI/dt slope represents the real inductance. The higher initial di/dt may be caused by diode reverse recovery current or by the specific inductor characteristic.

I'm under the impression that you are still trying to make the real circuit behave like the simulation, which is pointless in my view. It would be fine of course to understand all observed effects and reproduce them in the simulation, but if you don't manage it at the first attempt, you should take the circuit behaviour as is.

 

[adeel abid]

i have wound the 18 gauge wire on ferrite core & measured it LCR meter to be 300uH.
Is there any possibility that inductance value decreases as current increases due to core saturation or increase in temperature increases Rds(on) of IRF740 or am i missing some other parameter from device datasheet? i have simulated the circuit using orcad 9.2 pspice with IRF740 spice model. can't spice model not taking these secondary effects into account?
please advice.

thanks in advance

 

[BradtheRad]

During On-time 'the battery, inductor and IRF740 is in the circuit'.

Real hardware could add up to 2.5 ohms in real life. This is the amount which I found is able to reduce your current to 4 amps. When I first tried simulating your project, I deliberately added different amounts of resistance to find out what effect it had. (Post #2)

The simulator uses ideal components. It does not necessarily know how much resistance is in real components.

A real battery has internal resistance, a fraction of an ohm.
So does the coil.

If you look at mosfets, the specs say the On-resistance is incredibly low. However when I have used mosfets in real circuits, I never seem able to achieve the low amount stated in the specs. I didn't think I treated the mosfet so roughly that its operating characteristics were changed. And I thought I raised the gate voltage high enough, gave the gate a definite reference at the supply rail, etc.

 

[FvM]

All series resistance effects (inductor, Rdson, battery Ri) will show as an exponential distortion of the current ramp, they can't explain well the observed current waveform.

Core perrmeablity will usually increase from zero to medium current, but the effect would be neglectable for a gapped power inductor core. Other possible factors causing the unexpected current waveform can be identified by measurements.

 

[adeel abid]

what kind of measurements can be used to identify the real problem of adding resistance in circuit? i am more intrested in finding the faults/limitations that are making my circuit to depart from ideal behavior before going to adopt some other design.
i have plans to design a 220v 50hz inverter based on this converter. if i am not able to find my limitations/ faults of my circuit i may stuck at some later stage. i want to learn from my project.

please advice

thanks in advance

 

[BradtheRad]

what kind of measurements can be used to identify the real problem of adding resistance in circuit? i am more intrested in finding the faults/limitations that are making my circuit to depart from ideal behavior before going to adopt some other design.

The battery level will drop a little during On-time. Watch it on the oscilloscope. The amount of drop will indicate its internal resistance.

The coil emf may throw off the reading. It may work better to connect a 4-8 A ohmic load directly to the battery, and measure voltage.

Measure the coil's DC resistance directly.

While powering the circuit, check volt reading across clips, wiring, board traces, solder joints, etc. A high volt reading means high resistance. Even if all these things are in good condition, don't be surprised if they add up to a few tenths of an ohm.

The wiring that carries several amps should not be any longer than necessary.

Observe voltage across the mosfet in operation. The amount of drop will tell you its resistance during On-time. Again the coil emf may throw off volt readings. Perhaps instead of the coil you should insert a plain resistor (2 or 3 ohms).

If the gate voltage is too low, the mosfet will not reach its minimum On-resistance.

 

[adeel abid]

Please find the attached pictures of cores i am using to design inductor for boost converter. Please advice about its specs, material. whether it can give incorrect reading with my LCR meter its inductance increases as current through it increases? and whether it is suitable for boost converter application with Imax=8A, ton=190us and toff>=10us. if not suitable then which core should be used, advice about its physical appearence and other details. i am in doubt whether i am using the correct core for this application. color of core is green/blue ans yellow/white

thanks in advance